The Minecraft Geologic Survey includes data from 170,000 Minecraft worlds archived from four web sites between March and July 2014:

Planet Minecraft
The official Minecraft forum
Minecraft World Map
Minecraft World Share

Most of the actual worlds were downloaded from MediaFire and DropBox.

The full archive is about two terabytes in size, and includes worlds, screenshots, mods, texture packs, and skins. The Minecraft Geologic Survey exists to give a sample of the most interesting bits of the dataset, in a package that's small enough to reasonably distribute over the Internet.

The survey includes the following information for each world in the archive:

A constructed Minecraft world containing a "core sample" of the original world. The core sample is a semi-random sample of Minecraft chunks, 1% of the size of the original world, organized in a way to make programmatic access easy.

The sample is weighted to favor "interesting" chunks (those containing spawners, signs, and other entities). The sample will contain at least five chunks per dimension, and no more than 100. The sample also includes the world's "spawn chunk" (the chunk where a new player starts). The Reef maps are made out of spawn chunks.
The text and locations of all signs and books found in the world, in JSON format.
The commands used in each command block in the world, in JSON format.
Every in-game map found in the world, in the original binary map format.

For detailed information, and to download the MCGS dataset, see the README file.

plus-circle Add Review

comment

Reviews